Background: Current approaches to reverse HIV-1 latency use histone deacetylase inhibitors (HDACis) as a strategy to activate transcription from latent proviruses. A clinical trial involving administration of the HDACi panobinostat to HIV-1 infected individuals on long-term suppressive antiretroviral therapy (ART) demonstrated a measurable increase in HIV-1 transcription in CD4+ T-cells in blood. However, for effective viral clearance, panobinostat must activate transcription from a broad range of integrated proviruses. This study used single-proviral/genome sequencing to determine whether panobinostat selectively or nonselectively targets HIV-1 proviruses.
Methods: Peripheral-blood CD4+ T-cells were obtained from 15 participants before, during, and after panobinostat treatment. In addition, gut-derived CD4+ T-cells were obtained from a subset of participants before and during panobinostat treatment. Single-proviral/genome sequencing was used to characterise the genetic composition of the env region of cell-associated (CA) HIV-1 DNA and RNA to determine which HIV-1 proviruses were transcribed in response to panobinostat therapy within CD4+ T-cells. Additionally, we sequenced plasma HIV-1 RNA from samples collected during a post-HDACi analytical ART interruption. Maximum-likelihood trees were constructed for each participant and the average-pairwise distance of the sequences calculated using MEGA 6.0.
Results: The average-pairwise distance of the HIV-1 CA-RNA detected following administration of panobinostat was not significantly different from that of the HIV-1 CA-DNA (3% vs. 3%, p=0.862). Furthermore, upon phylogenetic analysis, the HIV-1 CA-RNA sequences intermingled with the HIV-1 CA-DNA sequences throughout the phylogenetic trees, supporting a broad and nonselective activation of HIV-1 proviruses. Plasma-derived sequences from the ART interruption samples contained expansions of identical sequences, which in three cases, were identical to HIV-1 CA-DNA sequences. Additionally, HIV-1 CA-DNA sequences from gut were genetically similar (≥99.7%) to those from the treatment interruption. Interestingly, peripheral blood HIV-1 CA-RNA had a significantly higher percentage of dead-end virus (hypermutated and/or containing stop codons) than the HIV-1 CA-DNA (24% RNA vs. 7% DNA, p=0.013), which has implications for the development of assays to predict the size of the latent reservoir.
Conclusions: Although a large amount of HIV-1 CA-RNA was replication incompetent, we identified CA-DNA from the peripheral blood that contributed to rebound virus during a post-HDACi ART interruption. In addition, we found HIV-1 sequences in gut-derived CD4+ T-cells that were similar to rebound virus during the analytical treatment interruption. These findings indicate that cells from both compartments, blood and gut, harbour replication competent virus that can contribute to viral rebound following treatment interruption and should therefore be targeted by eradication strategies. Importantly, we found thatpanobinostat nonselectively induces transcription from HIV-1 proviruses in HIV-infected individuals on long-term suppressive therapy which is promising for the development of future therapies that aim to activate quiescent HIV-1 proviruses as part of an eradication strategy.